Include Lagrange kernel trace polynomial term into DEEP composition polynomial

[Al-Kindi-0]

As the title says, this PR includes the missing term(s) in the DEEP composition polynomial related to the Lagrange kernel trace polynomial. A few remarks are in order:

1. Although we open at $z, z\cdot g, \cdots z\cdot g^{2^{\log(tracelen)-1}}$, we end up having to add only one term to the DEEP composition polynomial. This is because we use a multi-point quotient instead of single point quotients. This comes at a slight degradation of security which should be accounted for (in a future PR) in the security calculator.
2. During proof generation, the quotient from the previous point is currently computed using a naive polynomial division. This should be benchmarked to see if we might need to use an FFT-based division procedure.
3. Currently, the code makes an assumption (but doesn't enforce it) that the Lagrange kernel column is always the last one in the auxiliary trace. This makes the code for computing the DEEP polynomial cleaner.
4. Some TODO's are contingent on Generalize auxiliary trace building #271 .

[Al-Kindi-0]

Rebased and updated the code in light of #271

[irakliyk]

Looks good! Thank you! Not a full review yet - but I left some small comments/questions inline.

[irakliyk]

Looks good! Thank you! I left some more small comments inline.

Perhaps the biggest thing is the treatment of auxiliary trace width, related to this, the position of the Lagrange kernel column in the auxiliary trace. Basically, it is not immediately clear to me if we've correctly reduce the auxiliary trace width everywhere. I think the solution here would be to remove the ability to specify Lagrange kernel column index manually (and also handle construction of the Lagrange column within the library) - but this is for a different PR.

[irakliyk]

Looks good! Thank you! I left a couple of comments above. I think the main ones are:

• Move the new division benchmark into the math crate.
• Create an issue for investigating FFT-based division in the future.